|
Activity Number:
|
379
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #303540 |
|
Title:
|
Identifying Important Predictors Using L1 Penalized Models and Random Forests
|
|
Author(s):
|
Kellie J. Archer*+
|
|
Companies:
|
Virginia Commonwealth University
|
|
Address:
|
730 East Broad Street, Richmond, VA, 23298-0032,
|
|
Keywords:
|
random forest ; LASSO ; L1 penalization ; gene expression ; microarray ; machine learning
|
|
Abstract:
|
Microarray studies yield datasets consisting of a large number of candidate predictors for a small number of observations. Often it is of interest to develop a multigenic classi?er that is accurate and uncovers the predictive structure of the problem. Two more recently developed methods, L1 penalized models and random forests (RF), have good performance and yield estimates for each gene's contribution to the predictive structure. Moreover, both methods can be applied without reducing the dimensionality of the dataset. This study examined the e?ectiveness of L1 penalized models and random forest variable importance measures in identifying the true predictor among a large number of candidate predictors.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2009 program |
